Fragility Index Assessment of Randomised Trials Informing Endometriosis Management Guidance

Clinical practice guidelines for endometriosis rely on randomised controlled trials (RCTs) to inform surgical, medical, and postoperative management, many of which continue to inform contemporary care through ongoing clinical use and reaffirmation in guidance [1, 2]. The Fragility Index (FI) is a statistical tool that assesses the robustness of RCT results by quantifying how many outcome events would need to change to reverse statistical significance [3]. We aimed to evaluate the robustness of RCTs informing endometriosis management guidance by applying FI analysis. Given the existence of multiple national and international endometriosis guidelines, the ACOG Practice Bulletin was used as a representative case to examine the robustness of guideline-informing randomised evidence. We therefore reviewed all RCTs cited as level I evidence in this guideline [4]. Trials were eligible if they were randomised with surgically or clinically confirmed endometriosis, reported statistically significant, dichotomous primary outcomes that were explicitly dichotomous as defined by the original trial authors, and provided sufficient data for FI calculation. Exclusion criteria included non-randomised studies, single-arm trials, extension studies without a comparator arm, or trials in which endometriosis patients comprised only a subset not reported separately. Two reviewers independently extracted trial data. For each trial, the FI was calculated for the primary outcome reported by the authors, defined as the minimum number of event-status changes required to reverse statistical significance. The number of patients lost to follow-up was extracted as reported and compared with the FI. Secondary analyses compared FI values according to intervention type, comparator, and loss to follow-up. The Fragility Quotient (FQ) was calculated for each trial as the FI divided by the total sample size and summarised using median and interquartile range. Additional methodological details are provided in the Appendix S1. Of 27 RCTs cited as level I evidence, 16 met criteria for final analysis published between 1990 and 2010 (Table S1, Figure S1). Nine trials (56.3%) assessed surgery and seven (43.7%) assessed medical therapy. The overall median sample size was 81.5 (IQR 41.75–185.5), median FI was 3.75 (IQR 0.75–5.75), and the median number of patients lost to follow-up was 11.0 (IQR 1.25–21.25). Median FI was 3.5 (IQR 0.25–7.0) for medical and 4.0 (IQR 2.0–5.0) for surgical (p = > 0.999, Mann–Whitney U test). The number of patients lost to follow-up exceeded the FI in both groups (medical intervention—lost to follow-up = 22; surgical intervention—lost to follow-up = 11, Figure 1a). Trials with a placebo comparator demonstrated a higher median FI (5.0, IQR 3.5–8.0) compared with active comparator trials (1.5, IQR 0–3.5, p = 0.049, Figure 1b). In most placebo-comparator studies, the number of patients lost to follow-up outnumbered the FI (median FI = 5, lost to follow-up = 11). In the active-comparator group, the median number of patients lost to follow-up was similar to the median FI (median FI = 1.5, lost to follow-up = 0). The overall median FQ was 0.03 (IQR 0.01–0.08). Median FQ was 0.06 (IQR 0.02–0.09) for surgical trials and 0.02 (IQR 0.00–0.08) for medical trial. Additional analyses are provided in the Figures S2 and S3, illustrating FI values and patients lost to follow-up for individual trials. Our analysis underscores that RCTs informing endometriosis management guidance are characterised by low overall FI values, often exceeded by the number of patients lost to follow-up, highlighting sensitivity of statistically significant findings to small changes in outcomes and loss to follow-up. Robustness was similar for medical and surgical, although this comparison is limited by small sample sizes, but higher in placebo-controlled vs. active comparator trials. Interpretation of FI values should consider trial characteristics, particularly sample size and lost to follow-up, which substantially influenced robustness in this analysis [5]. Consistent with prior gynecologic surgery studies, low FI values were common despite statistical significance [6]. Low FI values may therefore partly reflect the small sample sizes typical of surgical and gynecologic randomised trials, rather than evidence weakness alone. In contrast, in other medical domains, FI analyses of guideline-referenced RCTs have demonstrated considerably higher FI values, particularly for surgical interventions [7]. This suggests that the evidence base for gynecologic surgery and endometriosis management may be uniquely vulnerable to the inherent limitations of trial design and outcomes. It is possible that large multicenter trials with larger sample sizes may increase the FI, as reported in oncology trials [7]. Guidelines require cautious use, transparent patient communication, and larger studies with fewer lost to follow-up to strengthen the evidence base. This analysis is limited by reliance on randomised trials cited in a major endometriosis guideline, many of which were published more than 15 years ago and may not fully reflect current practice. Nevertheless, these trials remain in ongoing use and are repeatedly reaffirmed in contemporary guidance, reflecting the persistent scarcity of large, high-quality randomised controlled evidence in endometriosis care and supporting cautious interpretation of guideline-informing evidence. Our findings highlight uncertainty and sensitivity to trial design, outcome selection, and loss to follow-up within guideline-informing evidence. Yael Yagur: conceptualization, study design, data extraction, manuscript drafting. Nir Horesh: data validation, manuscript editing. Gabriel Levin: manuscript editing. Raanan Meyer: conceptualization, study design, supervision, statistical analysis, manuscript editing. All authors approved the final version. The authors have nothing to report. R.M.—Intuitive Surgical, Claria Medical-consultant. All other authors report no conflicts of interest. This study is based on previously published randomised controlled trials cited in the ACOG Practice Bulletin on endometriosis. Table S1: Clinical trials included. Figure S1: Flow chart. Figure S2: Fragility Index values and number of patients lost to follow-up across randomised controlled trials of medical interventions for endometriosis. Figure S3: Fragility Index values and number of patients lost to follow-up across randomised controlled trials of surgical interventions for endometriosis. Please note: The publisher is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.